SIX CHAMBERS ✳ LOCAL-ONLY EXPERIMENTS ✳ NO TRACKING
◈
PERCEPTION ≠ RECORDING PLATE / 001
01 / THE GALLERY
HOW CERTAIN IS SEEING?
These illusions are controlled demonstrations, not tests of ability. Each stage begins with an editable drawing. Shift its parameters, switch to the reference state, or reveal its construction. The point is not to force an illusion: it is to learn where information ends and interpretation begins.
0 / 6 chambers explored
CHAMBER 01 / 06INTERACTIVE OPTICS
Ambiguous rotation
Two honest motions. One uncertain interpretation.
FIG 01 / RECONSTRUCTED VIEW
WHAT THE IMAGE IS DOING
A rotating pattern supplies insufficient depth cues. The direction you see can flip without the pixels changing.
WHAT THIS DOES NOT PROVE
The silhouettes rotate in screen space; there is no hidden three-dimensional object. Individual experiences vary; this page does not assess perception or health.
Inference note. A rotating pattern supplies insufficient depth cues. The direction you see can flip without the pixels changing.
CHAMBER 02 / 06INTERACTIVE OPTICS
The reversible room
When the vanishing point moves, the room changes its mind.
FIG 02 / RECONSTRUCTED VIEW
WHAT THE IMAGE IS DOING
Converging lines imply depth, yet a drawing is consistent with more than one spatial layout.
WHAT THIS DOES NOT PROVE
The apparent room is drawn from line intersections and a movable vanishing point. Individual experiences vary; this page does not assess perception or health.
Inference note. Converging lines imply depth, yet a drawing is consistent with more than one spatial layout.
CHAMBER 03 / 06INTERACTIVE OPTICS
Corridor of almost-infinity
Walk down a hallway that never existed.
FIG 03 / RECONSTRUCTED VIEW
WHAT THE IMAGE IS DOING
Repeated frames, scale and occlusion imply a tunnel. A flat surface can present depth-like structure.
WHAT THIS DOES NOT PROVE
Six nested quadrilateral frames create the impression of a receding corridor. Individual experiences vary; this page does not assess perception or health.
Inference note. Repeated frames, scale and occlusion imply a tunnel. A flat surface can present depth-like structure.
CHAMBER 04 / 06INTERACTIVE OPTICS
Colour is relational
The same colour, two different neighbours.
FIG 04 / RECONSTRUCTED VIEW
WHAT THE IMAGE IS DOING
Colour appearance depends on local context, adaptation and illumination assumptions. The two centre swatches here have identical RGB values.
WHAT THIS DOES NOT PROVE
Both swatches are exactly #aa83c7; only their surroundings differ. Individual experiences vary; this page does not assess perception or health.
Inference note. Colour appearance depends on local context, adaptation and illumination assumptions. The two centre swatches here have identical RGB values.
CHAMBER 05 / 06INTERACTIVE OPTICS
Parallax theatre
The foreground tells a different motion story.
FIG 05 / RECONSTRUCTED VIEW
WHAT THE IMAGE IS DOING
Relative motion can make a static scene appear layered. Stronger relative movement increases the depth interpretation.
WHAT THIS DOES NOT PROVE
Three flat layers move by different screen-space amounts; no 3D geometry is used. Individual experiences vary; this page does not assess perception or health.
Inference note. Relative motion can make a static scene appear layered. Stronger relative movement increases the depth interpretation.
CHAMBER 06 / 06INTERACTIVE OPTICS
The impossible staircase
Follow the path. Find the contradiction.
FIG 06 / RECONSTRUCTED VIEW
WHAT THE IMAGE IS DOING
An impossible object uses locally plausible junctions that cannot all hold in a single consistent 3D construction.
WHAT THIS DOES NOT PROVE
A stylised loop is assembled from inconsistent polygon joints; the paradox is pictorial, not physical. Individual experiences vary; this page does not assess perception or health.
Inference note. An impossible object uses locally plausible junctions that cannot all hold in a single consistent 3D construction.
02 / FIELD NOTES
THE WORLD IS NOT A SCREENSHOT.
Signals, hypotheses, and priors
The retina records patterns of light, not direct labels such as “corridor”, “shadow” or “object moving clockwise”. Perception integrates sensory evidence with expectations and context. In Bayesian models, a prior belief and sensory likelihood combine into a posterior interpretation; this is a modelling framework, not a claim that brains literally display probability equations.
Ambiguity is useful evidence
When an image supports several plausible scenes, the interpretation can switch. A reversal demonstrates that a percept is a construction constrained by input. It does not mean external reality is fictional, nor that every interpretation is equally reliable.
Limits of these chambers
The drawings simplify established illusion families rather than replicating validated laboratory stimuli. Monitor size, colour calibration, attention, vision differences and environmental light can alter what appears. A failure to experience an effect is an informative outcome, not a deficit.
Artistic metaphor / empirical boundary
Metaphor: “Reality glitches” describes the surprise of competing interpretations. Empirical claim: visual context and geometric cues affect perceptual judgments. These experiences cannot diagnose visual, neurological or psychological conditions.
03 / GUIDED EXPERIMENTS
TAKE THE IMAGE APART.
01 / CONTROL
Freeze the evidence
Choose rotation. Set a speed, pause, then reveal the mechanism. Does your perceived direction or depth change despite the same pattern?
02 / CONTEXT
Hold the centre constant
In Colour is Relational, drag contrast from minimum to maximum. Notice the centre squares keep the same hex colour. What actually changed?
03 / CONTRADICTION
Break the architecture
Manipulate the corridor and staircase. Which corners look locally plausible? Where does a single coherent 3D explanation stop working?